Bullying and Depressive Symptomatology Among Low-Income, African–American Youth
Utilizing a risk and protective factors approach, this research examined the relationship between self-reported depressive symptomatology, group membership (bully, victim, bully–victim) risks, and protection among a sample of African–American youths. Self-report data were collected in spring, 2002. Youth in grades 5–12 were sampled (n = 1,542; 51% female) from an urban school district in the Southeast. African–American youths self-identifying as bullies, victims, or bully–victims, reported higher levels of depressive symptoms compared to their nonbullied–nonvictimized counterparts. Additionally, multivariate results highlight a significant set of risk and protective factors associated with depressive symptomatology, even after controlling for the effects of self-identified group membership. These findings further contribute to our general understanding of the interplay among bullying, victimization, risk and protective factors, and their effects on depressive symptoms among a group of understudied African–American youth.
KeywordsBullyingVictimizationDepressive symptomsAfrican–American youth
Youth are more at risk for victimization than any other population subgroup in the United States; researchers have found American youth are 2.3 times more likely to be victimized than adults (e.g., Hanish and Guerra 2000). A recent national youth survey reported that 71% of sampled youth experienced some form of violent victimization in the past year (Finkelhor et al. 2005). One of the most common victimization experiences that are reported is being bullied. Several large school-age studies report that nearly one-quarter of youth are involved in bullying as a victim, perpetrator, or both (Juvonen et al. 2003; Nansel et al. 2001) with a growing literature continuing to show youth at greater risk for maladjustment because of their involvement in this form of aggression (e.g., Espleage and Swearer 2003; Hanish and Guerra 2002; Swearer et al. 2004). While a number of studies have examined the role of violence in determining mental health outcomes among African–American youth, we know much less about the psychosocial effects of this chronic, lower level form of aggression.
Bullying Among Youth
Historically, bullying has been characterized as normalized behavior typically found among adolescents. Yet, a comprehensive body of research clearly shows that bullying is a form of relational aggression that should not be viewed as acceptable or “normal” behavior among children and adolescents (e.g., Nansel et al. 2004). While there appear to be a variety of definitions, most research agrees that bullying is (1) chronic, (2) done with the intention to harm, (3) relational, and (4) is a form of exposure to violence that presents a major threat to healthy development through adolescence (Veenstra et al. 2005). It is in fact this last characteristic (threat to healthy development) that is important to the present study.
Victimization, Bullying, and Negative Outcomes Among Youth
Expanding on the pioneering work of Olweus (1980), researchers have identified unfavorable mental health outcomes associated with victimization due to bullying (Dupper and Meyer-Adams 2002; Hanish and Guerra 2000, 2002; Selekman and Vessey 2004). These and other researchers find that victims of bullying suffer a wide array of internalized and externalized maladjustments; victimized youths have lower self-esteem, difficulty in adjusting to the school setting, are more anxious, depressed, and withdrawn, and exhibit higher rates of avoidance behaviors than their non-victimized counterparts. Additionally, victims of bullying are reported to have higher rates of problem behaviors, such as drinking and smoking (Nansel et al. 2001). These unfavorable outcomes often persist into adulthood and affect the victims’ abilities to have healthy psychosocial outcomes (Dupper and Meyer-Adams 2002; Eisenberg and Aalsma 2005; Newman et al. 2005).
Likewise, studies have found elevated levels of depressive symptoms and other negative outcomes among school-aged youth who bully their peers (e.g., Eisenberg and Aalsma 2005; Fekkes et al. 2006; van der Wal et al. 2003). Bullies tend to be more aggressive, hostile, and uncooperative yet not as anxious or withdrawn as their victims or non-victimized counterparts (Veenstra et al. 2005); similar results are reported for youth who engage in bullying and also are victims of bullying by others (bully–victims). These aggressive victims actually exhibit high levels of both aggression and depression, and in some studies, appear to function more poorly than youth who are only victims or only perpetrators (Hanish and Guerra 2002; Veenstra et al. 2005). The present study examines these three categories more closely, their interplay with risk and protection and their relationship to depressive symptomatology among African–American adolescents.
Why only African–American youth? Although victimization of youth occurs regularly in American schools, the majority of studies evaluating the prevalence of youth victimization and mental health outcomes have primarily utilized samples of white, middle-class youth. As a result, the nature of bullying and the psychosocial outcomes for African–Americans and other victimized youth of color have gone largely understudied (Hanish and Guerra 2000; Newman et al. 2005; Storch et al. 2003). When African–American youth are included in the studies, conflicting results surface regarding the prevalence of victimization. While some research shows that African–American youth are less likely than both whites and Hispanics to be victimized, other studies indicate that African–American youth experience rates of victimization similar to those of whites (Graham and Juvonen 2002; Hanish and Guerra 2000; Nansel et al. 2001; Storch et al. 2003). However, few studies have contributed to a deeper understanding of the potential effects of victimization on racial/ethnic minorities and the extent to which African–American youth are adversely affected due to bullying. Additionally, few studies have looked at bullying among African–American youth only and thus are unable to eliminate the larger cultural and social dynamic that impacts interracial relationships.
There is also a lack of general knowledge regarding the psychosocial health outcomes of bullied African–American youth. Justification for using an African–American sample is provided by the arguments of Smokowski et al. (2004) who argue that using race comparative designs implicitly emphasize one cultural perspective over another. In addition, these models do not take into account that African–American youth face different cultural imperatives over the course of their development compared to Caucasian children. These race-comparative research designs may actually undermine our knowledge of adaptive development occurring within each specific group (Smokowski et al. 2004).
The current study, focusing only on African–American youths, is both timely and essential in order to further evaluate and better understand unique mental health outcomes for African–American youth. Understanding the complexity of African–American bullying and whether that experience is different from whites or other racial groups is beyond the scope of the current article. Nevertheless there have been some recent efforts to explore this phenomenon—asking if there is something unique about the African–American adolescent experience in the United States that might help to understand better the differences across racial groups if in fact they exist.
In a recent analysis of middle school bullying in the Southeast, Farrar (2006) employs a social constructionist perspective to understand both racial and gender differences in bullying. Specifically, the role of the stereotype is introduced as a cultural mechanism that adolescents are often forced to replicate or draw upon through a variety of mediums and, as a result, use it as a relational coping mechanism. Thus, the argument and the findings in this research show that Blacks bully more than Whites, suggesting that there must be something about the social and cultural meaning attached to being Black that corresponds with this type of externalizing behavior. If adolescents are limited in the ways they gain acceptance from others when using pro-social approaches, they may in fact turn toward more destructive, controlling bullying strategies (Farrar 2006; Unnever 2005). Certainly the argument has been made that minorities, particularly those growing up in low-income, poverty communities, are exposed to violence at exceptionally high rates in their families, schools, and neighborhoods. That exposure is often young adolescents viewing conflict and its resolution through violence. Would we expect to see similar outcomes (elevated bullying and victimization) among uni-racial samples of youth?
A Framework for Understanding Negative Outcomes Among Youth
Risk and protective factors research identifies domain specific factors that have significant effects on a number of internalizing behaviors among youth, specifically depression (e.g., Lyons et al. 2006; Saluja et al. 2004). Considerable research shows that risk and protective factors and their relationship to mental health outcomes among adolescents vary across gender, and age groups; females and older adolescents report more symptoms than males and younger adolescents (Sen 2004; Wild et al. 2004). While the present study is not examining in any detail demographic differences in depressive symptomatology, we control for these demographics and expect to find similar outcomes.
In addition to these important sociodemographic variables, our model proposes group membership (bully, victim, bully–victim) matters in explaining depressive symptoms among African–American adolescents. Research finds that depression is a common psychosocial outcome among adolescents who either perpetrate bullying or victimized by bullying (e.g., Dake et al. 2003; Fekkes et al. 2004). Compared to youth not involved in bullying behaviors, bullies, victims, and bully–victims were 2–6 times more likely to suffer symptoms of depression. Since the majority of these studies use White-Caucasian European or American samples, there still appears to be some question whether the relationship between bullying and depressive symptomatology is consistently significant among African–American youth. In addition, we examine the three distinct groups involved in bullying as they compare to youth not involved in bullying behavior. Prior research has clearly noted the important differences between the bully, victim, and the bully–victim (Juvonen et al. 2003; Unnever 2005). The depressive symptom differences among these three groups and the comparison group of youth who were not involved in bullying are examined both descriptively and in terms of how well group membership differences help to explain depressive symptomatology even after taking into account important risk and protective factors.
While risk factors are found in multiple domains (Hawkins et al. 1992), the present study focuses on three domains: individual, family, and school. We hypothesize that risk factors have direct and positive effects on depressive symptomatology among this sample of African–American youth. The first domain (individual) includes a factor assessing individual-level risk behaviors. Either individually or in a group, these variables represent an important collection of behaviors that have been consistently reported as being related to negative outcomes among youth generally, and specifically between different groups of youth involved in the bullying experience (e.g., Baldry and Farrington 2000; Dake et al. 2003). The variables used in the index created for this study closely mirror the Youth Risk Behavior Surveillance System (YRBSS) questions that assess risk behavior among high school youth in the United States (e.g., Brener et al. 2004).
The second domain (family) includes a factor assessing risk as it relates to the parent-child relationship. Research supports the use of family physical violence/abuse measures as a way to assess risk and its impact on mental health outcomes like depressive symptoms (Dube et al. 2001; Sofront et al. 2005). Physical or sexual abuse that takes place in the family setting is both an important risk factor for negative mental health outcomes as well as other behavioral outcomes among adolescents (Dearden et al. 2005).
A third domain (school) includes a factor that assesses school climate/school safety. School safety and feelings about being threatened have been noted as important determinants of mental health outcomes among youth (e.g., Hawkins et al. 1992). Previous work has noted the importance of the perception of school climate as an important risk factor in determining behavioral outcomes among youth (e.g., Fitzpatrick, 1999).
Likewise, the protective factors used in the model are also found in three domains: individual, family, and school. We hypothesize that they will have direct and negative effects on depressive symptomatology among this sample of African–American youth, even after controlling for bullying group membership. Protective factors are often considered important mediators—factors that represent more than just the absence of risk. These multi-domain factors operate to lower the negative effects of risk on health outcomes among youth. A growing literature is focusing on the role that protective factors play in the overall development of adolescents, partly because they can be manipulated in ways to incorporate them into prevention and intervention programming efforts (Piko et al. 2009; Van Voorhees et al. 2008).
The first domain of protection (individual) is a psychological resource, self-esteem. Self-esteem has been shown to be an important protection against negative mental health outcomes for youth. Several studies have demonstrated that feeling better about oneself has a lowering effect on mental health outcomes like depression, suicide ideation etc. (Fitzpatrick et al. 2005; Fitzpatrick et al. 2008; Salmon et al. 1998).
The second protection domain (family) focuses primarily on the supervisory role of the family but also includes the bonding role as well. For example, whether youth eat dinner with their parents or tell them where they are going, the protective mechanisms of the family are important to youth’s emotional stability. Prior research identifies them as important protective mechanisms when examining various negative outcomes among youth (Fitzpatrick et al. 2005; Wright and Fitzpatrick 2006).
The final domain of protection (school) focuses primarily on the importance of youth feeling as though they are connected/bonded to their school context. School bonding has been noted as a potentially important protective mechanism in reducing risk and negative health outcomes among youth (Ringeisen et al. 2003). Once again, research demonstrates this is a context where important intervention can take place; school matters in the psychosocial balance of youth.
Research Questions and Hypotheses
The present study is specifically interested in the bullying-depressive symptom relationship among African–American adolescents. Of particular interest to the present study are the following set of research questions: (1) Is there any difference in depressive symptomatology between youth depending on their level of involvement in bullying behavior (non-participants, bullies, victims, or bully–victims?; (2) What specific risk and protective factors are associated with depressive symptomatology among African–American youth?; and (3) Even after introducing risk and protection in the model, does group membership (bully, victim, bully-victim) still matter in explaining depressive symptoms among African–American youth? We explore these questions in an effort to further explicate the relationship between bullying group membership, risk, protection, and self-reported depressive symptomatology.
Prior research indicates significant differences in mental health outcomes among adolescents who engage in bullying behaviors (Dake et al. 2003). In particular, this research shows that mental health problems are greater among students who are involved in bullying as a perpetrator, victim or both, compared to youth who self-report not being involved in any aspect of the bullying experience. With a particular focus on depressive symptomatology, we examine this question of differences in an all African–American sample of youth and ask: Is there a significant difference in self-reported depressive symptomatology among these groups (non-involved, bully, victim, bully/victim) of students? Given the earlier review of an extensive bullying literature, we expect that victims will exhibit higher rates of depressive symptoms relative to the other groups, particularly the non-engaged/non-involved comparison group.
In the context of the mental health literature, two important demographic factors have been discussed as important correlates of depressive symptomatology: gender and age. While there are some subtle differences in their results, this research generally find that girls and older youth exhibit more depressive symptoms than boys or younger children. We have no reason to expect a different outcome in the present study and hypothesize that: older students and females will exhibit more depressive symptoms than younger, male students.
As discussed in detail earlier, we examine risk and protection and their role in understanding depressive symptomatology among African–American youth. Looking across multiple domains (individual, family, and school), what matters in predicting depressive symptom outcomes, controlling for bullying group membership? With regards to the earlier review of risk factors, we would expect to find that youth engaging in greater risk behaviors, being physically abused by their parent or guardian, and feeling unsafe in an uncertain school environment will report increased depressive symptomatology.
The role of protective factors in explaining depressive symptomatology is slightly more complicated. While we expect these factors to have a direct, negative effect on self-reported depressive symptoms, we also considered the role of these factors as mediators/buffers. When operating at elevated levels, does protection actually lower the overall negative effects of risk above and beyond their independent effect? Given the earlier discussion, we expect that youth with higher self-esteem, more family engagement in their lives, and higher bonding with their school/teachers will report fewer depressive symptoms than their counterparts. In addition, we examine the role of protective factors as mediators and/or buffers by examining both model shifts/changes (mediation) as well as testing for the presence of interaction between risk and protective factors (buffering).
Data and Methods
An urban school district with an overall school composition of more than 95% African–American was used for this study. Consent forms explaining the purpose of the health assessment were sent home to students in grades 5–12 by the school district office. If students opted out, they were asked to return a form letter of refusal signed by a parent or guardian (passive consent). The school district reviewed this protocol through their human subjects review panel and our university Institutional Review Board (IRB) also approved the general protocol. The survey took place during 1 week in May 2002. While we recognize that this form of consent is problematic, it was dictated by the school district. In addition to those students who returned forms not to participate, the remainder of non-respondents consisted of a combination of those not in attendance because of an excused or unexcused absence, suspension, or no longer attending school in the system (<10%).
Of the 2,464 eligible participants (based on average daily attendance) a total of 1,614 (65.5%) participated in the survey. Given the focus of the study, only surveys completed by African–American students were analyzed, reducing the number of survey respondents to 1,542, of which 51% were female. The final sample consisted of three age groups: 40.1% were from elementary school (grades 5 and 6), 26.4% from middle school (grades 7 and 8), and 33.5% were from high school (grades 9–12). The community of ~50,000 is located on the outskirts of a large Southeastern metropolitan area. The majority of residents living in the city are African–American (60%), with over one-third (35%) from female-headed households with children under the age of 18. In addition, nearly one-third of the residents were living at or below the poverty level; nearly 65% of the students were participating in the free or reduced lunch program in their schools.
The self-administered questionnaire was a health risk assessment administered by the school district consisting of 59 close-ended questions on sociodemographics, exposure to violence, risk-taking behaviors, and protective factors. The survey included a modified CES-Depression and self-esteem scales. Students were reminded of the confidential nature of their responses and that there were no unique identifiers that could link any of their answers to them personally.
Depression was operationalized using a shortened version of the original 20-item Center for Epidemiological Studies for Depression (CES-D) Scale designed to index affective depressive symptoms (Radloff 1977). The version used in the present study has eleven items and each one scored from 0 to 3 (M = 7.61, SD = 6.87). The eleven questions focus on the depressive affect domain of the CES-D. In addition to asking students whether they felt sad, lonely, or fearful, other questions asked students whether they had trouble getting along with others, felt hopeful about their future, and if they felt depressed. Respondents were asked to record the number and frequency of symptoms expressed. The categories included the following: 0 = <1 day, 1 = 1–2 days, 2 = 3–4 days, 3 = 5–7 days.
The formal properties of the original CES-D have been amply demonstrated with very high coefficients of internal consistency when the scale is administered to adolescents (Allgood-Merten et al. 1990; Avison and McAlpine 1992; Garrison et al. 1991). Among our sample of students, the brief CES-D was reliable, with a Cronbach’s alpha of .85. We weighted the brief CES-D for comparison purposes only, by a factor of 1.82 (the number of original CES-D items divided by the number of brief version items; i.e., 20/11 = 1.82). This weighting provides a rough comparison with other samples in which the full CES-D was used to assess adolescent depressive symptomatology. The new, weighted CES-D had a mean of 13.84 and a standard deviation of 12.51. This weighted version has been found to be significant in understanding other outcomes among African–American youth (Fitzpatrick et al. 2008; Dulin 2005).
Bullying Group Membership
In order to identify students who were bullies, victims, bully–victims, or not represented in any of these groups, respondents were asked how often in the last year they were bullied or bullied others in school. A detailed description of what bullying behavior meant was included in the question set for students. Students were instructed to consider a wide range of bullying behaviors including verbal, physical and emotional forms of bullying. The original bullying response categories reflected how many times they bullied someone in the past year including: 0 = 0, 1 = 1 or 2 times, and 2 = 3 or more times. Asking how often someone else bullied them at school assessed victimization during the past year. Those original response categories were 0 = 0, 1 = 1 or 2 times, and 2 = 3 or more times. Lastly, students identifying themselves as having been both victims and perpetrators of bullying during the past year were classified as bully–victims. The remaining students who responded zero (none) to both of these questions were classified as nonbullied/nonvictimized youth and served as the reference category throughout the analyses. The self-identified grouping variables (bully, victim, and bully/victim) were recoded into binary variables (1 = yes). This coding and variable construction closely follows other empirical work asking similar questions (e.g., Fekkes et al. 2004).
Two demographic variables, age (coded in years) and gender (1 = female), were included in the analysis. Both of these variables have been noted earlier to be important to understanding both bullying behavior and depressive symptom reporting.
As mentioned earlier, risk and protective factors operating in three domains—individual, family, and school—were identified as correlates of depressive symptomatology.
The first factor was an index of risk created from summed responses where students were asked how often they had been in a fight, a fight in school, carried a weapon, carried a weapon to school, suspended from school, received detention, and cut classes during school in the last 30 days? The individual risk behavior variables were coded with the following responses: 0 = never, 1 = one time, 2 = two or three times, 3 = four or five times, 4 = six or more times. The variable had a range of 0–16 (Mean = 6.7; SD = 6.2) and was reliable with a Cronbach’s alpha = .78. Higher values on the index indicate more risky and aggressive behavior.
The second variable assessed family risk, which asked respondents how often they had been slapped, punched, kicked or beaten by a parent/guardian in the last year. The categories were 0 = never, 1 = once or twice, 2 = often, 3 = all the time (Mean = .37; SD = .66).
The third factor was a school risk variable, which asked students whether they felt unsafe at school during the past year. The variable was coded as a dichotomy with 1 = felt unsafe (Mean = .46; SD = .69).
Protective factors within the three domains (individual, family, school) were also examined to see whether they had an independent effect on depressive symptomatology, and if as a group, had an impact on self-reported depressive symptomatology.
One protective factor identified within the individual domain was self-esteem. Self-esteem was measured using the Rosenberg (1965) 10-item Likert scale. Respondents were asked ten questions gauging their overall feelings of self-worth and self-satisfaction. Categories ranged from 1 = strongly agree to 4 = strongly disagree. The scores were summed and the final scale had a range from 4 to 40 (Mean = 23.0; SD = 3.5). The variable was reliable with a Cronbach’s alpha = .74.
In order to address the family domain, an index was created by summing responses from the following questions: “How often do you eat dinner with family members?”; “When you go out with your friends do your parents set a curfew?”; “When you go out with your friends, do your parents know where you are going?”; “When you come home from school, is there an adult at home waiting for you?”; How often do you talk with your parents about problems that are bothering you?” The five ordinal variables were coded as 0 = never, 1 = hardly ever, 2 = sometimes, 3 = most of the time and 4 = all of the time. The scores were summed and the final scale had a range from 5 to 20 (Mean = 13. 7; SD = 3.8). The variable was reliable with a Cronbach’s alpha = .74.
Finally, in assessing the school domain we asked respondents how much attention they received from their teachers during school. The response categories were 0 = never, 1 = hardly ever, 2 = some of the time, 3 = most of the time, and 4 = all of the time (Mean = 1.7; SD = 1.3).
Mean depression scores by bully and non-group membership (n = 1,524)
Regression estimates for sociodemographics, group membership, risk, and protective factors (n = 1,400)
Gender (female = 1)
School climate (unsafe = 1)
In the first model, gender and age are introduced. As predicted, age has a positive relationship with depressive symptoms (.098, p < .01); younger students report fewer symptoms compared to older ones. Additionally, as expected, gender is significant (.073, p < 01) with females registering nearly 2.5 points higher than males on the CES-D.
In the second model, bully (.077, p < .01), victim (.155, p < .01), and the bully–victim (.145, p < .01) groups (dummy-coded variables) all show significantly higher depressive symptoms relative to the reference group of non-bullied/non-victims. Bully–victims had the highest self-reported CES-D, which was more than six points higher than those who were neither bullies nor victims. Victims scored an average of three points higher on the CES-D compared to bullies.
In the third model, the risk factors from each domain (individual, family and school) were added and all three risks were in the expected (positive) direction and statistically significant. Students engaging in more risk-taking behavior (.163, p < .01), exposed to more family physical abuse (.125, p < .01), and report feeling unsafe in school (.169, p < .01), self-report more depressive symptomatology than their counterparts. Interestingly, when the block of risk factors were introduced into the equation, the bully group (.021, p > .05) and age (.038, p > .05) lost statistical significance. This may suggest that while bullying is important to determining ones’ psychosocial health, engaging in risk-taking behaviors or feeling unsafe in school or in the family may have a greater effect.
In the fourth and final model, protective factors were added. As a group of variables, they make a significant contribution to the explained variation in predicting self-reported depressive symptomatology; age (.014, p > .05) and bully (.017, p > .05) variables remained non-significant. Gender was significant (.143, p < .01) throughout the model changes confirming what we already know about gender being an important correlate of depressive symptoms among youth. There were no changes in the risk variables—all of the factors remained significant and in the hypothesized (positive) direction.
Self-esteem was significantly related (−.092, p < .01) to lower self-reporting of depressive symptoms among the students. Additionally, the family protection index was significant (−.077, p < .01) in reducing self-report symptomatology; the more respondents reported engaging with their family, the lower their depressive symptoms. While teacher attention was in the expected direction (negative) it did not achieve significance in the model (−.031, p > .05).
The total variance explained in the final model was approximately 15%. The R2 change was significant at the .001 level for each successive addition of the variable groups. Finally, while we might have anticipated that the protective factors would have had played some mediating role, we saw no preliminary evidence to suggest that role once protective factors were added into the regression equation. The overall equation remained stable; there were few changes in any of the coefficients and little reaction to group membership or risk variables with the addition of the protective factors. Likewise, after testing individual interaction effects among statistically significant risk and protective factors, we saw no evidence to suggest the presence of any buffering effect. While protection clearly matters, in this sample of youth, it appears to only minimally impact the negative effects of risk on depressive symptoms.
Overall, we believe this study significantly contributes to the literature on bullying and depressive symptomatology in several ways. First, this study highlights the general importance of evaluating bullying behaviors among a sample of African–American youth. This study also takes into account that African–American youth are at greater risk of exposure to violence and thus included factors associated with the violence perpetration and exposure to it (Fitzpatrick 1997). Results from this understudied population reveal that African–American youth engage in bullying behaviors in numbers similar to those of national estimates, thus supporting the suggestion of increased minority representation in future studies of bullying. These findings also suggest the importance of considering intra-racial bullying as an important issue that further questions the inter-racial findings in earlier work.
According to national estimates, approximately 15–20% of youth were victimized and 7–15% bullied others. While youth in this sample bullied and were victimized at rates similar to or slightly higher than national estimates, the research identified a unique subset of youth, bully–victims who comprised 10% of the group membership. Due to these distinctions among the participants, it is possible that bullying behaviors might be higher among this sample of youth when compared to national estimates.
Furthermore, the results indicate that among this sample of African–American youth, there were elevated levels of depressive symptomatology among all four self-identified student groups, including the comparison (non-victim/non-bully) group when compared to national benchmarks. Bullying behavior significantly affects the mental health of youth; all three “participating” groups were close to a CES-D score of 16, which is generally considered the clinical cut-off for further diagnostic consideration.
The relationship between gender and depressive symptoms for females in this sample finds they are susceptible to experiencing elevated levels of symptomatology despite bullying or being victimized. This finding is consistent with previous literature evaluating depressive symptom reporting among adolescents; being female is a risk factor associated with elevated depressive symptoms among youth (Farrar 2006; Lyons et al. 2006; Piko and Fitzpatrick 2003). Furthermore, research looking specifically into depressive symptomatology among African–American youth further substantiates that female gender is a risk factor for increased incidence of depressive symptoms among adolescents (Grant et al. 2004).
All of the risk factors were significant and operated in the hypothesized direction. As Smokowski et al. (2004) note in their research, it is essential to take culture into account when identifying risk and protective factors for children. What may be protective factors for some children might function as risks for other children. Clearly, additional research should attempt to identify and address the unique risk and protective factors associated with varying cultural experiences of different racial/ethnic groups.
Another significant risk factor for self-reported depressive symptomatology among this group of students was their perception of being in an unsafe school. Youth are keenly aware of their environments, and the perceived dangers in the school hinder their development and impede their normal day-to-day interactions. A stronger, safer support network must be developed within the school system. In order to develop this network, input from school personnel, parents, and children should be solicited.
As noted earlier, physical violence/abuse within the home was a significant risk factor in explaining depressive symptomatology. Youth who reported corporal punishment in the home were more likely to report more depressive symptoms than youth who were not corporally punished. The psychosocial risks associated with corporal punishment have been well documented and appear to function in similar manners among this sample of African–American youth (Ohene et al. 2006).
One protective factor, self-esteem, was instrumental in reducing depressive symptom reporting for all groups. This underscores the importance of psychological resources in establishing resiliency in the face of risk. Certainly one method for combating this increased depressive symptom reporting among the victim and bully–victim groups might be to find ways for them to improve their self-perceptions.
Psychosocial research has consistently demonstrated that not every person with equivalent roles exposed to similar stressors share comparable experiences. There are protective factors, which help reduce the intensity of the stressors. Using this perspective adds to our understanding of the continuing importance and contribution of the risk and protective factors approach to our understanding of mental health outcomes. Accounting for risk and protective factors associated with depressive symptoms shows how the stress of bullying and the associated risks are related to protective factors. Thus, the perspective proposed in this research is supported by the results—bullying is a stressor that affects psychosocial well-being, specifically among victims and bully–victims.
While informative, this research has some limitations. First, the data used in this research are cross-sectional and thus we are unable to establish causality. The results obtained from this research can only highlight associations or describe events. In addition, we recognize that the causal link between self-reported depressive symptoms and certain risk and protective factors is tenuous. For example, it is difficult to establish the ordering of the relationships between depressive symptomatology and bullying group membership. Elevated levels of depressive symptoms may put one at risk for victimization (bullying); school bullies are able to prey on particular types of students and mental health complications may make these students more susceptible to violent outcomes. Nevertheless, the use of multivariate techniques to conduct this research does allow for an estimation of both individual and block effects of the independent variables, something which is clearly important to understanding these complex relationships and design issues for future studies.
Several caveats are necessary concerning the use of the brief CES-D. First, because the measure only assesses the frequency of symptoms, not the severity, we consider our sample of students to be on some continuum of depressive symptom reporting rather than in specific categories that correspond to degrees of clinical depression. Second, our measure is a self-reported assessment of feelings that students reported over a 2-week period. Thus, we are not in a position to report clinical diagnoses. In addition, we acknowledge the weakness of self-reported assessments generally in determining mental health status, particularly among adolescents. Finally, although the full CES-D was the most desirable to use in the current study, we were somewhat constrained by the length of the questionnaire and the single class period of approximately 45 min. Other studies facing time constraints have adopted brief versions of the CES-D and have had considerable success, yielding valid and reliable responses. It should be noted, though, that, in the majority of cases, those responses were from adult populations (e.g., Kohout et al. 1993; Melchoir et al. 1993).
Additionally, the non-response rate, while not excessive, may lead to bias. Results were not obtained for students who did not fully complete the survey, from students who were absent or suspended for administrative reasons, in alternative schools, or who did not receive parental permission. It is possible that, if these students were included in the results, bullying behaviors might actually be higher than what was estimated by this sample. Students are also asked to recall events that occurred over the past year, which may also lead to recall bias.
It is important to note that the results from this study can only be generalized to African–American children in grades 5–12 with similar characteristics. Additional studies are needed to further confirm these findings. The goal of this study was an evaluation of behaviors for low-income African–American students in one urban school district in the Southeast. Therefore, any policy implications or interventions should be uniquely tailored to this particular group or groups with similar characteristics.
Another limitation of this work is the lack of data on sibling violence within the home. It is possible that this form of victimization may be associated with high levels of depression among this sample of youth. While work on sibling violence has been conducted outside the US, with results substantiating a relationship between sibling violence and bullying at school, more work in this area needs to be conducted (Wolke and Samara 2004).
Evaluating the risk and protective factors associated with depressive symptoms among the specific subgroups of youth (bullies, victims, bully/victims) has clearly helped to disentangle the effects of bullying behavior on the mental health outcomes of an understudied population, African–American youth. This research highlights the importance of constructing safe environments for children to allow for unfettered psychosocial development. Family violence and children feeling unsafe at school significantly impede their psychological development. Specifically, any interventions geared toward these factors must be uniquely tailored to African–American youth, taking into account community and familial factors that might contribute to the proliferation of risks. Programming should also take into account the needs of the target community as well as input from community members and youth.
By “removing” race/ethnicity from the current analysis, we examined intra-racial behavior among youth in a way unlike the majority of bullying research has over the past several decades. Even recent work (Vervoot et al. 2008) continues to struggle with understanding the inter-racial/inter-ethnic dynamic in schools and its role in determining bullying behavior. Does heterogeneity matter? Classroom and school composition clearly are important factors in determining both positive and negative relational outcomes among students across racial and ethnic groups. Yet, it is very difficult to measure the impact of underlying cultural or structural arrangements present in schools or communities. How much does community inter-racial/ethnic history impact the current relationships among different groups of youth in their schools? Is this artifact critical to determining present behavior and does it represent an important element to consider when designing bullying prevention material?
Our findings reinforce the importance of examining the bullying-depressive symptomatology relationship among African–American youth. First, the relationship varied across bullying groups—while victims and victim–bullies self-reported more depressive symptoms than the non-bullied or non-victimized groups, bullies did not. Even when risk and protective factors were introduced, these groups still reported more depressive symptoms than their counterparts. Clearly, experiencing this type of chronic violence, either as a passive or aggressive victim, has a significant impact on these student’s stress levels. Furthermore, our findings point out that these bullying groups are unique both in terms of who more likely engages in bullying or more likely to be bullied, as well as what effects these experiences have on their mental health. Second, we noted earlier that depressive symptom reporting was higher among females, yet once bullying group membership and other variables were introduced, the expected “age effect” disappeared. The developmental nuance of experiencing and coping with extraordinary stressors is well documented particularly among older female adolescents who generally report more symptomatology. Nevertheless our results suggest something slightly different and may require a closer examination of these relationships, particularly when using minority school-age samples. Finally, the findings indirectly suggest the need for carefully designed prevention and intervention programming. The interplay between chronic violence and mental health problems continues to go unnoticed, particularly among low-income, minority school populations. Culturally sensitive, age-appropriate, and multi-faceted programs that recognize the link between exposure to violence and depressive symptoms seem important, especially because of their reported links to school absence, poor grade performance, dropping out of school, and other negative outcomes that have implications into adulthood.